Many implantable devices that interact with tissue, including those used in surgical procedures, in-vitro tests, and in-vivo implantations, require special care for accurate positioning (location and orientation) of the implantation device. Furthermore, a critical issue is to ensure that implantation occurs satisfactorily; that is, the device is inserted in at the required depth without device failure. Manual insertions of devices cannot provide this level of control in positioning and insertion, therefore leading to high rate of device failure during insertion, over-design of devices with larger-than-needed foreign materials, and functional failures. An important need is to have automated mechanisms for insertion, that provide precision in positioning (cellular-scale, approximately 20 μm), orientation (±0.5), and speed control (±1%), as well as allow feedback and evaluation through visual and sensor-based in-situ characterization capability.
An illustrative example of this need arises from the insertion of the neural probes for brain-computer interfaces (BCI). Research on BCI and brain-machine interfaces (BMI) in recent years has demonstrated the feasibility of driving motor prostheses for the upper limbs of amputees and for restoring mobility to quadriplegics and tetraplegics whose condition arose due to injury or disease. More recently, research has begun to focus on providing feedback loops between the brain and other nervous tissue and the computers and machines to which they are interfaced by stimulating the tissue with signals from the external equipment to return sensation to BMI and BCI recipients. In this way, an injured or diseased individual can control an external prosthetic and receive sensation from it in a way that naturalistically mimics the limb they lost or the biological function that is impaired.
BCIs and BMIs comprise: 1. an interface to the soft tissue that records the electrical, chemical or mechanical activity of the soft tissue and transduces it to a signal in a suitable energy domain, typically electrical, 2. a decoder that extracts the information from the signals received from the tissue, 3. a transmitter that sends out the decoded signals, 4. a receiver of the decoded signal, 5. a computer or machine that acts under the instructions carried in the decoded signals, 6. a sensor array that detects changes in the environment caused by the action of the computer or the machine and transduces it to a signal in a suitable energy domain, 7. an encoder that receives the output of the sensor array and converts it to a sensory signal for transmission, 8. a transmitter that sends out the encoded sensory signals, 9. a receiver of the encoded sensory signals, and 10. an interface that transduces the encoded sensory signals to an electrical, chemical or mechanical signal for stimulation of the soft tissue in which the interface is embedded.
The interface is a critical feature of BMIs and BCIs and its placement must be as close as possible to the biological signal sources without damaging them in order to maximize the information extracted from the soft tissue and minimize the amount of energy needed to transmit sensory information back into the soft tissue. The most common interface is the electrode. Typically, this is an insulated, electrically conductive material with a small surface exposed to the soft tissue environment. Electrodes have dimensions ranging from 10s of micrometers to 100s of micrometers. The effectiveness, stability and reliability of these interfaces has been identified in the literature, in part, as dependent on the method of implantation and the accuracy of their placement. Interface reliability is a critical research area where progress is needed prior to transitioning BMI and BCI technology for practical restoration of motor and sensory functions in humans. Two key issues are 1) the inability of current interfaces to reliably obtain accurate information from tissue over a period of decades, and 2) currently measured signals from tissue cannot be reliably used to control high degree-of-freedom (DOF) prostheses with high speed and resolution.
Failure of biological soft tissue interfaces may be caused by several issues. After implantation, current probes are surrounded by reactive microglia and reactive astrocyte scarring as shown pictorially in FIG. 1(a). In the brain, damage to the neural vasculature causes a breach in the blood-brain barrier (BBB) that is associated with reactive soft tissue responses. Tissue reaction with the probe results in encapsulation that insulates the electrode by impeding diffusion of chemical and ionic species and may impede current flow from the soft tissue to the interfaces. Encapsulation increases the distance of the electrode from active neurons. For viable recording, the distance of the electrode from active neurons must be less than 100 μm. Progressive death and degeneration of neurons in the zone around the inserted probe due to chronic inflammation may eliminate neural electrophysiological activity. Lastly, interconnects may fatigue and break due to stresses. Experiments in animals have resulted in some neural electrode sites failing while others keep working for several years. This variability in outcome is believed to be due to several factors including variable BBB damage, variable scar formation, mechanical strain from micromotion, inflammation, microglial condition and disconnected neurons.
Tissue interfaces employed today for BMI and BCI applications come in a variety of shapes made of many materials and apparatus and methods for implanting these interfaces must have the functional and design flexibility to handle the multiplicity of devices available today and accommodate the designs and forms that become dominant as the technology matures and moves into widespread human use. In the next few paragraphs, the challenge presented by the range of device types and materials will be established by reviewing the devices described in the literature.
Historically, the interfaces have been stiff needles usually made from wires, silicon or glass. Metal wire neural probes are typically 50-100 μm in diameter and usually made of platinum or iridium and insulated with glass, Teflon, polyimide or parylene.
Silicon-mounted interfaces made with MEMS fabrication were first introduced by Ken Wise and Jim Angell at Stanford in 1969. Ken Wise's group at the University of Michigan subsequently developed a series of silicon probes and probe arrays with multi-site electrodes.
A 2D probe array was developed at the University of Utah in 1991, known as the Utah Electrode Array (UEA). The UEA has become a favored interface in human applications in the central nervous system (CNS) and for research in the peripheral nervous system (PNS).
Polycrystalline diamond (poly-C) probes with 3 μm thick undoped poly-C on a ˜1 μm SiO2 layer have been fabricated by Dr. Aslam's group at Michigan State University.
Research groups have created more compliant probes made with thin-film wiring embedded in polymer insulating films. Flexible CNS probes have been made in polyimide, SU8/parylene and all parylene. These probes are still extremely stiff in both axial and transverse directions relative to brain tissue, which has a Young's modulus of approximately 30 kPa. Any axial force transmitted through the external cabling directly acts on the probe and creates shear forces at the electrode-tissue interfaces. Such forces may come from external motion or from tissue growth around the implant. To address this issue, a group from Carnegie Mellon University and the University of Pittsburgh have developed a parylene-coated Pt probe with a thickness of 2.5 μm and width 10 μm that provides axial strain relief in the brain through a meandered design (FIG. 1(b)). The cables external to the brain are also meandered to further reduce transmission of brain-skull relative motion to the embedded probe. Because of the size and compliance of the meandered probes they are embedded in a biodissolvable delivery vehicle which provides the stiff structure for implantation.
A team from Drexel Univ., the Univ. of Kentucky and SUNY created ceramic-based multisite microelectrode arrays on alumina substrates with thickness ranging from 38 to 50 μm, platinum recording sites of 22 μm×80 μm, and insulation using 0.1 μm ion-beam assisted deposition of alumina.
Y.-C. Tai's group at Caltech produced parylene-coated silicon probes with integral parylene cabling, shown in FIG. 2(a). The shanks were up to 12 mm long. A primary innovation was a flexible 10 μm-thick, 830 μm-wide, 2.5 mm-long parylene cable.
Flexible polyimide probe arrays (FIG. 2(c)) have been made with gold electrodes. These probes must be inserted by first creating an insertion hole with a scalpel or needle. A later polyimide probe array incorporated silicon for selected locations along the length of the shank, with polyimide connectors to create enhanced compliance, as shown in FIG. 2(b).
An innovative all-polymer probe design incorporated a lateral lattice-like parylene structure attached to a larger SU8 shank to reduce the structural size close to the electrodes. The lattice structure, shown in FIG. 2(e), included a 4 μm-wide, 5 μm-thick lateral beam located parallel to the main shank. Encapsulating cell density around the lateral beam was reduced by one-third relative to the larger shank. While the structure was non-functional, it is presumed that placing electrode sites on the smaller beam would result in superior recording performance.
U.S. patent application 20090099441 from Dr. Giszter's Drexel group describes biodegradable stiffening wires 1 braided with electrode wires 2 (see FIG. 2(f)) where flexible wires 2 are braided onto a maypole structure 4 with stiff biodegradable strands 1. When the biodegradable strands 1 dissolve, the flexible wiring 2 is left in the brain tissue. These braided composite electrodes are similar in spirit to present invention. However, reliable and manufacturable connections to the braided wires become difficult when scaled to arrays.
Olbricht et al has reported on flexible microfluidic devices supported by biodegradable insertion scaffolds for convection-enhanced neural drug delivery. The device consists of a flexible parylene-C microfluidic channel that is supported during its insertion into tissue by a biodegradable poly(DL-lactide-co-glycolide) (PLGA) scaffold. The scaffold is made separately by hot embossing the PLGA material into a mold.
Tyler et al, have developed a neural probe made from a polymer nanocomposite of poly(vinyl acetate) (PVAc) and tunicate whiskers, inspired by the sea cucumber dermis. The probe material exhibits a real part of the elastic modulus (tensile storage modulus) of 5 GPa after fabrication. When exposed to physiological fluid conditions, its modulus decreases to 12 MPa.
The trend in devices is towards more compliant materials and structures that will have stringent implantation requirements in terms of speed, force and placement. In the following paragraphs, the state-of-the-art in soft tissue interface insertion technology is described.
Manual implantation or stereotaxic assisted implantation by a skilled surgeon is the most common method of implantation of the variety of interfaces and interface delivery vehicles described above. Manual implantation means the procedure is done by hand and stereotaxic assisted implantation means it is done through the use of a stereotaxic frame that holds the interface delivery vehicle and provides a hand operated screwdrive to position and insert the interface. Positioning is usually performed with the assistance of a stereomicroscope that provides some measure of depth perception. With this technique, there is no control over the speed of insertion and only gross sensitivity to the profile of the underlying soft tissue, both of which could contribute to the variability observed in the outcomes of soft tissue interface implantations. Insertions of the Michigan probe array are done using this method.
The low velocity of manual insertions, either by hand or using stereotaxic frames, results in observable soft tissue dimpling prior to penetration of the tissue. Dimpling was found to be accompanied by soft tissue compression that resulted in damage to the tissue and reduced signal extraction.
To improve outcomes by reducing manual variability and increasing insertion speed, research groups adopted a hand-held pneumatic insertion device invented by Normann et al. and experimentally demonstrated by Rousche and Normann. The pneumatic inserter has a piston mechanism that is actuated pneumatically to strike an endpiece rod on which is adhered the device to be implanted. The burst of pressure accelerates the piston and its momentum is transferred to the endpiece rod which is driven toward the brain at speeds of 8 m/s, which was found to be required for the 10 electrode×1 electrode interface to penetrate the soft tissue. An adverse effect of the mechanism is recoil of the endpiece due to the return spring which can lead to retraction of the interface device if it remains adhered to the endpiece. Researchers using the UEA avoid this effect by resting the interface on the tissue into which it will be implanted and using the endpiece to strike the back of the interface device. This technique does not allow for accurate placement of the interface in soft tissue because there is no visibility of the contact points between the interface and the tissue. House et al. achieved a measure of control over the spatial relationship between the endpiece and the device to be inserted by mounting the pneumatic inserter on a stereotaxic manipulator. They found the impact between the endpiece and the backside of the interface often led to damage of the interface, so they added a “footplate” to the device. However, because the device is not mechanically connected to a fixed reference structure, it is subject to elastic recoil from the soft tissue into which it is implanted and this can lead to retraction of the interface from the tissue. To overcome retraction the interface must be over-driven into the soft tissue so that after recoil, the full length of the interface remains in the tissue. The literature does not have detailed studies on the impact of over-driving the insertion on the health of the recipient.
The literature reports other insertion mechanisms of varying levels of complexity and functionality. Rennaker et al. reported a manually positioned spring-driven hammer mechanism for insertions up to 1.5 m/s for microwires mounted on the insertion device using a locking screw. Jensen et al. reported a hydraulically driven micromanipulator with manual positioning and force sensing and a speed of 2 mm/s. Dimaio and Salcudean used a robotic manipulator to implant 17 gauge epidural needles with force sensing but did not report the insertion speeds they achieved. Bjornsson et al. used stepper motors to implant Si microneedles at up to 2 mm/s with force sensing. Sharp et al. electronically controlled a micromanipulator with an in-line load cell to achieve insertion speeds from 11 μm/s to 822 μm/s for evaluation of penetration mechanics in cerebral cortex. In each of these cases and others in the literature, fine positioning, if done, was performed by visually locating the interface over the soft tissue to be implanted.
Accurate placement of the interface requires referencing of the tissue height, maintaining the relative height between the interface mounted on the insertion apparatus and the tissue as the tissue surface moves under pulsatile and respiratory motion, mapping and identifying the insertion location with an overlay of the interfaces and positioning the interface in space with respect to the tissue surface. This is an area where the literature is very sparse. Kozai et al. used two photon imaging to map the cortical vasculature to identify target locations prior to interface implantation and found that when this is done the trauma of implantation can be reduced by 73% for surface vasculature compared to the case when vasculature is targeted.